Project Summary Epithelial ovarian cancer (EOC) affects approximally 21,000 women a year in the USA resulting in 13,000 deaths. Standard treatment includes debulking surgery followed by adjuvant chemotherapy. For 80% of women this treatment is effective and prolongs survival. However, in 20% of women the cancer is extensively disseminated through the peritoneum at time of surgery which complicates the surgical procedure and does not allow optimal tumor debulking. For these women, tumor debulking is not effective and they experience complicated and prolonged postoperative recovery. A recent randomized phase III trial demonstrated that neoadjuvant chemotherapy with interval debulking surgery is an effective alternative treatment for ovarian cancer patients and may be the ideal approach for patients who cannot undergo optimal up front debulking. Thus there is a need to identify and stratify patients based on their response to debulking surgery and develop more effective surgical and chemotherapeutic approaches targeting sub-optimally debulked tumors To address this need, we performed a meta-analysis of gene expression data using publicly available profiles of 1,525 ovarian cancers and identified 198 genes that were highly expressed in tumors that were not optimally debulked. We refer to these genes as ?debulking signature. Ontologic pathway analysis of the debulking signature showed hyper-activation of a specific oncogenic signaling responsible for malignant cancer behaviors such as dissemination resistance to chemotherapy, i.e. the TGF-? pathway. Thus, the signature may serve as a predictive biomarker for patients who would benefit from up-front surgery and provide a biological rationale for novel targeted therapies of tumors that cannot be optimally debulked. The goal of this project is to develop a validated genomic signature which can be developed into clinical diagnosis, and test in ovarian cancer mouse models whether targeting one of the most enriched pathways of this signature, TGF-?, is effective. We will validate the 198 genes identified as highly expressed in EOC that are not optimally debulked using two independent tissue arrays and establish an optimal genomic signature that can be used for pre-operative diagnosis of these tumors (aim 1). We will then perform preclinical studies testing whether inhibitors of the TGF-? pathway currently being used in clinical trials for other cancers, improve management of disseminated ovarian cancer models in mice (aim 2). Altogether, we will establish a predictive biomarker that assists the surgeon and patient to choose the best surgical procedure to be applied to an EOC patient, as well as identify a new adjuvant chemotherapeutic option that improves therapeutic outcome. If successful, these studies will spare women from therapeutic suffering and prolong their lives.